Vapi says its voice agents now handle about a billion calls a year

Why it matters

Vapi is selling the infrastructure beneath enterprise voice agents. Its reported call volume shows deployment at scale, while the unresolved measure is whether those calls produce reliable outcomes and justify expanding automation.

Y Combinator said in a September 25th post on X that Vapi's voice-agent platform handles about a billion calls a year for companies including Amazon, Uber and Intuit. The figure puts the startup's pitch in production terms: its software is being used to handle phone interactions at enterprise scale, not just to generate a convincing demo.

The annual figure needs a clear time frame. In May, Vapi said it had passed 1 billion calls made across the platform, a cumulative milestone. TechCrunch reported that CEO Jordan Dearsley put Vapi's then-current volume at between 1 million and 5 million calls per day. Annualized, that wide daily range runs from roughly 365 million to 1.8 billion calls. It makes the September post's approximate billion-a-year claim plausible, but neither the X post nor the company's funding announcement provides an independently audited call count or defines precisely how the annual figure is calculated.

Dearsley and co-founder Nikhil Gupta arrived at Vapi after building Superpowered, a productivity startup that went through Y Combinator. The founders had spent years working together; Vapi's origin came when Dearsley built a voice-based AI therapist for his daily walks in 2023. The therapy product failed to find a market, but the low-latency voice infrastructure behind it attracted developer interest. The founders shifted toward selling that infrastructure, and Vapi launched publicly in 2024, according to TechCrunch's account of the company's history.

That pivot explains the product's position in the market. Vapi sells a developer platform for building, deploying and managing voice agents, giving customers control over components such as speech recognition, language models and generated speech. Its documentation describes integrations with phone systems and customer workflows, including support and appointment scheduling. The company is competing to become infrastructure for businesses that want to put AI into existing phone operations, rather than selling only a prebuilt receptionist or call-center bot.

Amazon Ring is the strongest example of that infrastructure bet in public reporting. TechCrunch said Ring evaluated more than 40 vendors before choosing Vapi and now routes all its inbound calls through the platform. Ring's software-development vice president, Jason Mitura, told Vapi's May funding announcement that the deployment moved from zero to production in two weeks, and that Ring's customer-satisfaction scores improved. Those are customer-reported outcomes, not a published measurement of how many calls Vapi resolved without human help or what the deployment saved Ring.

The distinction matters commercially. Call volume establishes that systems are being used; it does not by itself show how many conversations they complete successfully, how often they hand off to people, or whether customers prefer them. Those operating measures will determine whether Vapi's enterprise deployments expand beyond high-volume support queues. For businesses buying the software, control over model choice, latency, escalation and monitoring may matter as much as a headline call count.

Vapi's May financing gives investors a stake in that infrastructure layer. The company announced a $50 million Series B led by Peak XV, with M12, Microsoft's venture fund, Kleiner Perkins, Bessemer Venture Partners and earlier investors participating. The announcement put total funding at $72 million. TechCrunch reported the round valued Vapi at about $500 million after investment and said an investor source described revenue as a healthy eight-figure annual recurring run rate; Vapi's release separately claimed enterprise ARR had grown tenfold, without specifying the comparison period. The revenue description and growth figure remain attributed claims, rather than audited financial disclosures.

The customer list in the September post includes Amazon, Uber and Intuit. Vapi's May release named Amazon Ring, New York Life, Intuit, Kavak and ServiceTitan among enterprise customers. Those deployments support the case that the platform has moved into real business workflows. They do not establish that each named company uses Vapi for the same type or volume of calls, or that all use is at the same production scale.

The roughly billion-calls-a-year figure is therefore best read as a marker of throughput and distribution. Vapi's founders started with a voice application that did not catch on, then built the plumbing that other companies could use to deploy their own agents. Investors have now funded that platform with a $50 million round; the harder test is whether the calls it carries produce reliable outcomes customers will keep paying to scale.